Synopsis Three risk of collision incidents, involving four aircraft, occurred within a period of about two minutes. A Fairchild Metro III, on an instrument flight rules (IFR) flight arriving from the southwest and cleared for a visual approach to runway 31, passed with 0.7 mile horizontal and 500 feet vertical separation behind a Fokker F-28, which was one of two aircraft also on visual approach to runway 31. About one minute later, while in a left turn to rejoin the approach to runway 31, the Metro passed with 0.4 mile horizontal and no vertical separation to the left of a Cessna 152, the pilot of which was on a visual flight rules (VFR) flight conducting traffic observations east of the approach to runway 31. About 30 seconds later, the F-28 was cleared to land on runway 31 when a Cessna 414 was holding position for take-off on the runway. Shortly thereafter, the F-28 initiated a missed approach. The Board determined that the first risk of collision incident occurred because the crew of the Fairchild Metro III misidentified the preceding aircraft on approach. Contributing factors were the confusing visual cues in the approach environment, the lack of separation standards other than visual separation for IFR aircraft on visual approaches, and the arrival controller's reluctance to intervene. Possible contributing factors were the lack of an approach sequence number, the reduced staffing level at the Area Control Centre, and that the annual refresher training for the controllers had not been carried out. The second risk of collision incident occurred when the crew of the Fairchild Metro III altered course to the east of the approach path to increase separation from the F-28. Contributing factors were the airport controller's delay in issuing instructions to the crew of the Fairchild Metro III to turn, and the Cessna 152 pilot's arrangement with the Tower to turn down the volume on his radio. The third risk of collision incident occurred because the airport controller overlooked the Cessna 414 that was holding position for take-off on runway 31 when he cleared the F-28 to land. Contributing to the occurrence was the airport controller's preoccupation with the first and second risk of collision incidents. 1.0 Factual Information 1.1 History of Flight This report pertains to three risk of collision incidents, which will be referred to as the first risk of collision (ROC)(1), second ROC, and third ROC respectively. The three ROCs occurred within the three minutes and 50 seconds that elapsed from the time Bearskin Lake Air Service flight 404 (BLS 404), a Fairchild Metro III, received its clearance for a visual approach, until the time Canadian Regional Airlines flight 1328 (TAF 1328), a Fokker F-28, initiated a missed approach. The three occurrences took place at latitude 4954'N and longitude 9715'W, during the hours of daylight and approximately a half-hour before the official sunset for the day, which was at 1705 central standard time (CST). 1.1.1 History of Flight - First Risk of Collision The Fairchild Metro III (BLS 404) arrived in the Winnipeg terminal control area on an instrument flight rules (IFR) flight from Brandon, Manitoba, to the Winnipeg International Airport, Manitoba. Air Canada flight 178 (ACA 178), a McDonnell-Douglas DC-9, was the first aircraft in sequence on approach for runway 31, followed by the Fokker F-28 (TAF 1328). At 1617:31 CST, the Fairchild Metro III crew first called the arrival controller 37 nautical miles (nm) southwest of Winnipeg, at 15,200 feet above sea level (asl)(2). At 1623:25 CST, at an altitude of 7,000 feet asl, the crew(3) reported that the airfield was in sight, and the arrival controller advised them to follow the F-28 traffic turning final just through their twelve o'clock position and about 8 nm, down low. The F-28 referred to was TAF 1328, whose altitude at the time was 3,800 feet asl. The Fairchild Metro III crew observed one aircraft on approach, at their 11:30 position(4), that fit the description of the F-28. They advised the arrival controller that they had the traffic in sight, and were cleared for the left hand visual approach to runway 31 to follow the F-28. The crew carried out the aircraft landing checks after receiving approach clearance, and were reportedly aware that traffic information was based on their ground track. The arrival controller did not provide the Fairchild Metro III crew with a sequence number, nor was he required by regulation to do so. The crew did not request one. At 1624:41 CST, to draw the crew's attention to the fact that their aircraft was rapidly closing with the F-28, the arrival controller advised them that their groundspeed was 290 knots and that the F-28's ground speed was 160 knots. The aircraft that the crew was observing was on short final approach at a position left of 12 o'clock, and they believed they had ample spacing behind the traffic. The crew acknowledged the call and took no further action. The arrival controller was reportedly reluctant to interfere in the conduct of visual approaches because crews are responsible for maintaining their own separation with the aircraft they are following. At 1625:49 CST, the arrival controller asked the Fairchild Metro III crew whether the F-28 was still in sight, and the crew responded that it was. The arrival controller then informed the Fairchild Metro III crew that their groundspeed was 70 knots greater than that of the F-28 which was at their 12 o'clock and at a distance of one mile(5). The Fairchild Metro III crew acknowledged this, re-directed their attention to this last position and immediately saw the F-28. The Fairchild Metro III crew started a right turn to increase separation between the two aircraft, and passed through the runway 31 approach path .7 nm behind the F-28, with 700 feet of vertical separation. 1.1.2 History of Flight - Second ROC The Cessna 152 (C-FGEK) had departed from the Winnipeg/St. Andrews airport about 50 minutes earlier and was on a visual flight rules (VFR) flight plan. The pilot's traffic watch duties consisted of observing street traffic in various parts of the City of Winnipeg, and periodically broadcasting reports to a public radio station with a portable radio. The Cessna 152 pilot advised the airport controller at 1622:16 CST that he would be going off Tower frequency in one minute to make several street traffic reports. The airport controller had approved, with the proviso that the Cessna 152 stay in the area between the approaches to runways 25 and 31. As the Fairchild Metro III was passing behind the F-28 and through the approach path for runway 31, Arrival instructed the Fairchild Metro III crew to call Winnipeg Tower on their frequency. The crew called the airport controller at 1626:05 CST, and asked whether he wanted the flight to use runway 25. The airport controller discussed this request with the arrival controller and, at 1626:26 CST, advised the Fairchild Metro III crew of the Cessna 152 traffic watch aircraft, and instructed the crew to start a left hand 360-degree turn to rejoin the approach to runway 31. The Fairchild Metro III crew confirmed that the traffic was in sight and started the turn. Radar data shows the Cessna 152 northwest bound, turning left, at the time of the closest proximity between the Fairchild Metro III and the Cessna 152, while the Fairchild Metro III was northbound, and also turning left(6). During the turn, the Fairchild Metro III passed the Cessna 152 with .4 nm horizontal and no vertical separation. At 1626:50 CST, the airport controller broadcast an instruction to the Cessna 152 to turn north; however, no reply was received. The Cessna 152 pilot's next call to Tower was at 1628:54 CST, when he advised that he was again monitoring the frequency. 1.1.3 History of Flight - Third ROC After the DC-9 landed, the pilot of C-FSAL, a Cessna 414, advised Tower that he was ready for take-off, and received clearance at 1625:48 CST to move to position and hold on runway 31. The Tower voice recording shows that the airport controller was then in almost continuous communication with the arrival controller and the crews involved in the second ROC, until the F-28 approached the runway(7). At 1627:00 CST, the F-28 crew asked whether there was an aircraft on the runway, but this transmission was almost unintelligible because it coincided with a transmission between Tower and Arrival concerning the second ROC. The F-28 crew then received clearance to land on runway 31. Shortly thereafter, at 1627:10 CST, the F-28 crew advised Tower that they were carrying out a missed approach. The crew reported that they initiated the missed approach at about one-half mile from the runway from an altitude of about 300 feet above ground level (agl), when they visually confirmed that the Cessna 414 was on runway 31. 1.2 Injuries to Persons 1.2.1 Injuries to Persons - Fairchild Metro III (BLS 404) 1.2.2 Injuries to Persons - F-28 (TAF 1328) 1.2.3 Injuries to Persons - Cessna 152 (C-FGEK) 1.2.4 Injuries to Persons - Cessna 414 (C-FSAL) 1.3 Damage to Aircraft There was no reported damage to any of the aircraft involved. 1.4 Other Damage 1.5 Personnel Information All flight crew members were certified and qualified for their respective flights in accordance with existing regulations. There was no evidence that incapacitation or physiological factors affected the performance of any of the flight crew members. Both air traffic controllers were certified and qualified in accordance with existing regulations. 1.5.1 Personnel - Fairchild Metro III (BLS 404) The Fairchild Metro III crew started their duty day in Brandon, Manitoba at 0900 CST and had flown about three hours before the occurrence. After their arrival in Winnipeg, scheduled for about 1630 CST, the crew planned to carry out a return flight to Brandon at 1740 CST and complete their duties for the day at about 1900 CST. The first officer had flown eight hours in the previous three days, and 80 hours in the previous 30 days. The captain had flown 7.5 hours in the previous three days, and 70 hours in the previous 30 days. The crew's previous duty day ended at 2215 CST. Transport Canada regulations limit flight crew duty days to 15 hours, and limit flight hours to 120 in the previous 30 days, and 300 in the previous 90 days. 1.5.2 Personnel - F-28 (TAF 1328) 1.5.3 Personnel - Cessna 152 (C-FGEK) 1.5.4 Personnel - Cessna 414 (C-FSAL) 1.5.5 Air Traffic Controller - Arrival 1.5.6 Air Traffic Controller - Airport 1.6 Aircraft Information 1.6.1 Aircraft Information - Fairchild Metro III (BLS 404) 1.6.2 Aircraft Information - F-28 (TAF 1328) 1.6.3 Aircraft Information - Cessna 152 (C-FGEK) 1.6.4 Aircraft Information - Cessna 414 (C-FSAL) 1.6.5 DC-9 and F-28 Configuration and Paint Scheme The DC-9 (ACA 178) and the F-28 (TAF 1328) are both twin-engine turbojet, T-tail configured aircraft with rear fuselage-mounted engines, and are of a similar size. The DC-9's fuselage was painted mostly white, and the tail was mostly dark green. The F-28's fuselage was blue on the lower half and light grey on the upper half, and the tail was painted blue. 1.7 Meteorological Information The surface weather report for the airport at 1600 CST was as follows: 4,000 feet scattered, visibility 15 miles, temperature four degrees Celsius, dew point minus three degrees Celsius, winds 280 degrees at 20 knots, altimeter setting 29.68 inches of mercury. The scattered layer of cloud was reported as strato-cumulus covering 3/10 of the sky. In the weather report at 1700 CST, the ceiling was listed as 3,500 feet broken, visibility 15 miles, and the broken layer of cloud was reported as strato-cumulus covering 7/10 of the sky. The forecast upper winds for Winnipeg at 3,000 feet asl were 270 degrees true at 24 knots; at 6,000 feet they were 260 degrees true at 29 knots. 1.8 Aids to Navigation Runway 31 is served by an instrument landing system (ILS), which was reportedly operating normally at the time of the occurrences. The final approach track is 313 degrees magnetic (M). The Winnipeg Air Traffic Control (ATC) Radar Modernization Project (RAMP) radar system was commissioned in February 1993, and was the sole radar system in use at both Winnipeg Terminal and Winnipeg Tower during the occurrences. The radar system was reportedly operating normally throughout the time of the occurrences. 1.9 Communications Aircraft approaching Winnipeg while operating under IFR are initially controlled by Winnipeg Centre during en route flight. Air traffic control of incoming flights is then handed over to Winnipeg Arrival control (Arrival) on frequency 119.5 megahertz (MHz), and then to Winnipeg airport control (Tower) on frequency 118.3 MHz. Winnipeg Tower is also able to receive and transmit on 125.4 MHz. 1.10 Aerodrome and Airspace Information The Winnipeg Area Control Centre (ACC) provides air traffic control services for airspace in the vicinity of Winnipeg. One of the units within the ACC, the Terminal specialty (TCU), provides air traffic control for the airspace at Flight Level (FL) 240 and below, within a 35 nm radius of the Winnipeg Airport, as well as the area between the 230 and 305 degree radials of the Winnipeg very high frequency omni-directional range (VOR) transmitter, between 35 and 55 nm at 6,000 feet and below. The Terminal specialty is further divided into Arrival, Departure, and Terminal Radar Service Advisory (TRSA) positions. The TRSA is a part-time VFR radar advisory position which is combined with the Departure position during periods of low traffic levels. The Winnipeg airport control zone extends 7 nm from the centre of the airport, from the ground to 3,000 feet asl. The control zone is categorized as Class D airspace; flights within it must establish two-way communication with the appropriate ATC facility unless prior approval is received from ATC. 1.11 ATC - Arrival A Transport Canada Resource Alignment Project was carried out in 1993, and a Transport Canada Unit Evaluation was conducted from 19 to 29 September 1994. The Resource Alignment Project results, as adjusted by the Unit Evaluation, were that 27 controllers were required on staff for the Winnipeg Terminal specialty. At the time of the occurrences, 22 licensed controllers were on staff, of whom 16 were available for work. The Arrival position was staffed by one controller, who was assisted at the time of the occurrences by another controller in the Terminal Data (Data) position. The data controller, who was also qualified at the Arrival position, managed the flight data board, updated data strips, and performed other related tasks. The level of traffic at the time of the occurrences was light to moderate, with normal complexity. The arrival controller was a team leader (supervisor) in the Terminal specialty, and worked a 36- day schedule consisting of three sets of five days on, four days off, and one set of six days on and three days off. At the time of the occurrences, he was working his fourth day after one day off, and had worked 11 of the previous 12 days, and 25 of the previous 30 days, including seven overtime shifts. The controller's supervisory duties include stand-back supervision of air traffic control services in the Terminal specialty, preparing the work schedules of other controllers and carrying out performance reviews and appraisals of other controllers. The arrival controller was working a regular controller position; a shortage of available controllers reportedly prevented him from conducting stand-back supervision. The controller also reportedly had difficulty in finding time to complete his supervisory tasks on schedule. The controller reported that his on-the-job efficiency may have been affected by the amount of time he was required to spend at work lately. Fatigue is subjective in nature and each individual varies in his or her ability to maintain stability and control under conditions which are generally regarded as conducive to fatigue, such as poor health, inadequate sleep or rest, excessive working hours, and stressful or physically tiring work. Studies have shown that excessive controller workload resulting from understaffing presents a potential for controller fatigue(8), and several previous safety recommendations have been issued as a result. 1.12 ATC - Tower The airport controller completed his qualification training in the Winnipeg Tower in July 1993, after transferring from a controller position in Thunder Bay. At the time of the occurrences, the airport controller was working his second day after three days off. He had worked 8 of the previous 14 days, and 16 of the previous 30 days. He was reportedly well rested at the time of the occurrences. The level of traffic at the time of the occurrences was considered moderate. Tower communications congestion was moderate initially, increasing to heavy as the occurrences progressed. 1.13 ATC Recurrent Training The Air Traffic Services Administration and Management Manual (ATSAMM) provides that operational personnel shall demonstrate that their performance of operational duties meets unit proficiency standards on an ongoing basis. This performance is assessed by a Performance Review and Assessment of Potential (PRAP). In addition, annual refresher training is to be provided to all operational personnel (Section 551). Critical tasks that are not performed on a regular basis are to be included in the refresher training. In addition, new topics and topics of interest are brought up for discussion as they arise. The last refresher training session provided to the arrival controller was completed 07 May 1993. Since that date, two routine reviews were carried out that indicated that the arrival controller was able to perform independently at all times, and with superior skill in many instances. The last refresher training course at the Tower was held in February and March, 1992. The airport controller received no refresher training between the time of his qualification as an airport controller at the Tower and the time of the occurrences. However, a routine review was carried out that indicated that the Tower controller was able to perform independently at all times. 1.14 Approach Control Procedures and Separation Criteria 1.14.1 Terminal Area Approach control procedures are governed by the Transport Canada Air Traffic Control Manual of Operations (MANOPS). MANOPS specifies that aircraft in the terminal area operating under IFR in the circumstances of the occurrence aircraft shall have at least three miles horizontal or 1,000 feet vertical separation. 1.14.2 Visual Approaches Visual approaches may be requested by pilots or initiated by controllers to gain an operational advantage. MANOPS section 547.1(b) provides that controllers may clear an IFR aircraft for a visual approach if the crew of the aircraft being vectored is instructed to maintain visual separation from preceding aircraft arrivals. MANOPS section 547.1(c) provides that an approach clearance is not issued until the crew reports sighting the aircraft that they will be instructed to follow and from which they will be maintaining visual separation. Once the aircraft is cleared for the visual approach, the arrival controller is no longer responsible for maintaining separation from the preceding aircraft; therefore, no minimum separation standards (e.g. altitude or distance) apply, other than the visual separation maintained by the crew. Controllers often provide approach sequence numbers to aircraft on visual approaches, and point out their traffic using the o'clock system, although they are not required to do so. The arrival and data controllers reported that they controlled Fairchild Metro aircraft differently than other aircraft. Crews flying Metros reportedly fly faster and steeper approaches than those in other types of aircraft, and they prefer visual approaches to avoid delays resulting from ATC guidance. The Approach controller offered the Fairchild Metro III (BLS 404) crew a visual approach without waiting for a request from the crew. 1.14.3 Terminal/Tower Inter-Unit Agreement The current inter-unit agreement between the Winnipeg Tower and the Winnipeg ACC establishes procedures for the handling of air traffic between Tower and TCU. It provides that Tower shall operate within the airport control zone and TCU shall operate outside the control zone, and: D.1.5: Communications and control transfer between Tower and TCU positions shall not be effected until pertinent conflictions at the transferring position have been resolved; and D.1.6: Arrival shall provide minimum separation involving VFR aircraft as follows: a. same runway - 2 nm. The conflict that arose from the first ROC had not been resolved when Arrival passed control of the Fairchild Metro III (BLS 404) to the airport controller on Tower frequency as that flight was about 1 nm behind and closing with the F-28 (TAF 1328). 1.15 Recognition of Visual Targets The Winnipeg airport is located in the northwest part of the City of Winnipeg, at an elevation of 783 feet asl. Aircraft crews arriving from the west and planning to use runway 31 will see aircraft already on approach for that runway against a backdrop of the buildings (lighted at night) of the City of Winnipeg, when the preceding aircraft are at a lower altitude. Studies have shown that aircraft viewed against an urban background are unusually difficult to recognize.(9) Empirical studies have been carried out to determine the ability of human observers to see and recognize visual target forms at various distances. One such study(10) measured the time required by observers with normal vision to recognize targets of various angular sizes, and their error rate in doing so. It found that ...when the maximum dimension of the target is less than 12 minutes of visual angle (MOA), there is a precipitous rise in both the search time and errors, and concluded that: It may be presumed that 12 minutes of angle is the lower limit that is acceptable for target recognition for the type of forms used in this study.(11) Controllers reportedly do not receive training concerning human physiology as it relates to target recognition. The aircraft that the Fairchild Metro III crew had initially seen and had been following was the DC-9 (ACA 178). When the Fairchild Metro III (BLS 404) crew received approach clearance, the F-28 (TAF 1328) subtended 4.5 MOA, at a range of 11.5 nm in the 12:30 position relative to the Fairchild Metro III. The DC-9 subtended 6.2 MOA at a range of 9.7 nm in the 11:30 position relative to the Fairchild Metro III. When the arrival controller advised the Fairchild Metro III crew of the relative speed difference between the F-28 and themselves, the DC-9 subtended 8.6 MOA, and the F-28 subtended 11.8 MOA. The Fairchild Metro III crew was at this time watching the DC-9, and the F-28 was 55 degrees to the right of the DC-9.